273 views

Uploaded by vvigya

- Face Recognition using Transform Domain Techniques
- sigma-delta
- EC1306 Digital Signal Processing Laboratory [REC]
- signal processing toolbox in matlab
- SPARSE NONNEGATIVE MATRIX BASED ON -DIVERGENCE FOR SINGLE CHANNEL SEPARATION IN COCHLEAGRAM
- Noise Cancellation In Speech Signal Processing Using Adaptive Algorithm
- Electrical Power Systems Electrical Power Engineering Power Systems
- fpga new manual.docx
- Modern Techniques for Protecting and Monitoring of Transmission Lines
- PSO_2012.pdf
- Lab S5 DLTI GUI and Nulling Filters
- 330401696
- ECE-320-Spring-2004
- Hiring Needs Survey-Sept08
- fir response
- Application Form for Navy
- Embeded 7
- 05578688
- Ch10_FIr_ Filter Design_structure_Lecture.ppt
- lecture 1 dsp

You are on page 1of 6

Matthew P. Donadio

m.p.donadio@ieee.org

18 July 2000

1 Introduction

As data converters become faster and faster, the application of narrow-band extraction from

wideband sources, and narrow-band construction of wideband signals is becoming more

important. These functions require two basic signal processing procedures: decimation and

interpolation. And while digital hardware is becoming faster, there is still the need for

efficient solutions. Techniques found in [CR83] work very well in practice, but large rate

changes require very narrow band filters. Large rate changes require fast multipliers and

very long filters. This can end up being the largest bottleneck in a DSP system.

In [Hog81], an efficient way of perfoming decimation and interpolation was introduced.

Hogenauer devised a flexible, multiplier-free filter suitable for hardware implementation, that

can also handle arbitrary and large rate changes. These are known as cascaded integrator-

comb filters, or CIC filters for short.

This paper sumarizes the findings published in [Hog81]. An overview can also be found in

[Fre94]. An extension of CIC filters has been published in [KJW97], and is briefly mentioned

here. When in doubt, the reader should refer to these sources.

2 Building Blocks

The two basic building blocks of a CIC filter are an integrator and a comb. An integrator is

simply a single-pole IIR filter with a unity feedback coeficient:

This system is also known as an accumulator. The transfer function for an integrator on the

z-plane is

1

HI (z) = (2)

1 − z −1

1

Using the equations from [OS89] for a single pole system, we can determine that

1

|HI (ejω )|2 = 2(1−cos ω)

−1 sin ω

ARG[HI (ejω )] = −tan

1−cos ω (3)

jω undefined ω = 0

grd[HI (e )] =

− 21 ω 6= 0

The power response is basically a low-pass filter with a −20 dB per decade (−6 dB per

octave) rolloff, but with infinite gain at DC. This is due to the single pole at z = 1; the

output can grow without bound for a bounded input. In other words, a single integrator by

itself is unstable.

z −1

#?

- -

"!

A comb filter running at the high sampling rate, fs , for a rate change of R is an odd-

symetric FIR filter described by

In this equation, M is a design parameter and is called the differential delay. M can be any

positive integer, but it is usally limited to 1 or 2. The corresponding transfer at fs

ARG[HC (e )] = − RM

jω

2

ω

(6)

grd[HC (ejω )] = RM

2

When R = 1 and M = 1, the power response is a high-pass function with 20 dB per decade

(6 dB per octave) gain (after all, it is the inverse of an integrator). When RM 6= 1, then the

power response takes on the familiar raised cosine form with RM cycles from 0 to 2π.

When we build a CIC filter, we cascade, or chain output to input, N integrator sections

together with N comb sections. This filter would be fine, but we can simplify it by combining

it with the rate changer. Using a technique for multirate analysis of LTI systems from [CR83],

we can “push” the comb sections through the rate changer, and have them become

2

at the slower sampling rate fRs . We accomplish three things here. First, we have slowed down

half of the filter and therefore increased efficiency. Second, we have reduced the number of

delay elements needed in the comb sections. Third, and most important, the integrator and

comb structure are now independent of the rate change. This means we can design a CIC

filter with a programmable rate change and keep the same filtering structure.

- z −M

#?

- -

"!

fs , followed by a rate change by a factor R, followed by N cascaded comb stages running

at fRs . A CIC interpolator would be N cascaded comb stages running at fRs , followed be a

zero-stuffer, followed by N cascaded integrator stages running at fs .

#

- I - I - I - R

?

- C - C - C -

"!

#

- C - C - C R

- 6 - I - I - I -

"!

3 Frequency Characteristics

The transfer function for a CIC filter at fs is

RM −1

!N

−RM N

(1 − z ) X

H(z) = HIN (z)HCN (z) = −1 N

= z −k (8)

(1 − z ) k=0

This equation shows that even though a CIC has integrators in it, which by themselves

have an infinite impulse response, a CIC filter is equivalent to N FIR filters, each having a

3

rectangular impulse response. Since all of the coeficients of these FIR filters are unity, and

therefore symetric, a CIC filter also has a linear phase response and constant group delay.

The magnitude response at the output of the filter can be shown to be

sin πM f N

|H(f )| = (9)

πf

sin R

By using the relation sin x ≈ x for small x and some algebra, we can approximate this

function for large R as

N

sin πM f 1

|H(f )| ≈ RM

for 0 ≤ f < (10)

πM f M

We can notice a few things about the response. One is that the output spectrum has nulls

at multiples of f = M1 . In addition, the region around the null is where aliasing/imaging

occurs. If we define fc to be the cutoff of the usable passband, then the aliasing/imaging

regions are at

(i − fc ) ≤ f ≤ (i + fc ) (11)

for f ≤ 21 and i = 1, 2, · · · , b R2 c. If fc ≤ M2 , then the maximum of these will occur at the

lower edge of the first band, 1 − fc . The system designer must take this into consideration,

and adjust R, M , and N as needed.

Another thing we can notice is that the passband attenuation is a function of the number

of stages. As a result, while increasing the number of stages improves the imaging/alias

rejection, it also increases the passband “droop.” We can also see that the DC gain of the

filter is a function of the rate change.

4 Bit Growth

For CIC decimators, the gain G at the output of the final comb section is

G = (RM )N (12)

Assuming two’s complement arithmetic, we can use this result to calculate the number

of bits required for the last comb due to bit growth. If Bin is the number of input bits, then

the number of output bits, Bout , is

It also turn out that Bout bits are needed for each integrator and comb stage. The input

needs to be sign extended to Bout bits, but LSB’s can either be truncated or rounded at

later stages. The analysis of this is beyond the scope of this tutorial, but is fully described

in [Hog81].

For a CIC interpolator, the gain, G, at the ith stage is

i

2 i = 1, 2, · · · , N

Gi = 22N −i (RM )i−N (14)

R

, i = N + 1, · · · , 2N

4

As a result the register width, Wi , at ith stage is

and

WN = Bin + N − 1 (16)

if M = 1. Rounding or truncation cannot be used in CIC interpolators, except for the result,

becuase the small errors introduced by rounding or truncation can grow without bound in

the integrator sections.

It is now worth revisiting the unstable aspect of the integrator stages. It turns out that

it is not a problem. For decimators, integrator overflow is not a problem as long as two’s

complement math is used and we don’t expect an overall system gain > 1. For interpolators,

the comb stages and zero stuffing will prevent integrator overflow.

5 Implementation Details

Because of the passband droop, and therefore narrow usable passband, many CIC designs

utilize an additional FIR filter at the low sampling rate. This filter will equalize the passband

droop and perform a low rate change, usually by a factor of two to eight.

In many CIC designs, the rate change R is programmable. Since the bit growth is a

function of the rate change, the filter must be designed to handle both the largest and

smallest rate changes. The largest rate change will dictate the total bit width of the stages,

and the smallest rate change will determine how many bits need to be kept in the final stage.

In many designs, the output stage is followed by a shift register that selects the proper bits

for transfer to the final output register. A system designer can use the equation for Bout

for a decimator and W2N for an interpolator to calculate proper shift values.

For a CIC decimator1 , the normalized gain at the output of the last comb is given by

(RM )N

g= (17)

2dN log2 RM e

This lies in the interval ( 12 , 1]. Note that when R is a power of two, the gain is unity. This

gain can be used to calculate a scale factor, s, to apply to the final shifted output.

2dN log2 RM e

s= (18)

(RM )N

which lies in the interval [1, 2). By doing this, the CIC decimation filter can have unity DC

gain.

Filter sharpening can be used to improve the response of a CIC filter. This technique

applies the same filter several times to an input to improve both passband and stopband

1

This paragraph is an generalization of equations found in the datasheet for the Harris/Intersil HSP50016.

5

charecteristics. If H(z) is a symetric FIR filter, then a sharpened version, HS (z), can be

expressed as

The magnitude response of a sharpened CIC filter would then be

!2N !3N

sin πM f sin πM f

|H(f )| = 3 − 2 (20)

sin πf

R

sin πf

R

The interested reader is referred to [KJW97] for more details. Please note that it uses

different parameters and implements a CIC filter a bit differently than [Hog81].

7 Conclusion

Since their inception, CIC filters have become an important building block for DSP systems.

They have found a particular niche in digital transmitters and receivers. They are currently

used in highly integrated chips from Intersil, Graychip, Analog Devices, as well as other

manufacturers and custom designs. This paper has attempted to summarize key points

found in [Hog81] and provide some insight into designs. While many journal submissions

are of limited value to an engineer, this paper was written for designers. As such, the reader

should try to locate [Hog81] as the definitive reference for CIC filters.

References

[CR83] Ronald E. Crochiere and Lawrence R. Rabiner. Multirate Digital Signal Processing.

Pretice-Hall Signal Processing Series. Prentice Hall, Englewood Cliffs, 1983.

Academic Publishers, Boston, 1994.

[Hog81] E. B. Hogenauer. An economical class of digital filters for decimation and inter-

polation. IEEE Transactions on Acoustics, Speech and Signal Processing, ASSP-

29(2):155–162, 1981.

[KJW97] Alan Y. Kwentus, Zhongnong Jiang, and Alan N. Wilson, Jr. Application of filter

sharpening to cascaded integrator-comb decimation filters. IEEE Transactions on

Signal Processing, 45(2):457–467, 1997.

Pretice-Hall Signal Processing Series. Pretice-Hall, Englewood Cliffs, 1989.

- Face Recognition using Transform Domain TechniquesUploaded byIOSRjournal
- sigma-deltaUploaded byabhingp01
- EC1306 Digital Signal Processing Laboratory [REC]Uploaded byjeyaganesh
- signal processing toolbox in matlabUploaded byLộc Vĩnh Đào
- SPARSE NONNEGATIVE MATRIX BASED ON -DIVERGENCE FOR SINGLE CHANNEL SEPARATION IN COCHLEAGRAMUploaded byTJPRC Publications
- Noise Cancellation In Speech Signal Processing Using Adaptive AlgorithmUploaded byEditor IJRITCC
- Electrical Power Systems Electrical Power Engineering Power SystemsUploaded byHariprasath Selvakumar
- fpga new manual.docxUploaded byshubham soni
- Modern Techniques for Protecting and Monitoring of Transmission LinesUploaded bySrinivasan Srini
- PSO_2012.pdfUploaded bymenguemengue
- Lab S5 DLTI GUI and Nulling FiltersUploaded byArmando Cajahuaringa
- 330401696Uploaded byimrancola5191
- ECE-320-Spring-2004Uploaded byNitin Soni
- Hiring Needs Survey-Sept08Uploaded byAlex
- fir responseUploaded byRavi Tiwari
- Application Form for NavyUploaded byNaveen Kumar
- Embeded 7Uploaded byahamed
- 05578688Uploaded byAyan
- Ch10_FIr_ Filter Design_structure_Lecture.pptUploaded byPraneeth Kumar
- lecture 1 dspUploaded bywahbaabass
- Non Lin Filt Neuro Morph NetworksUploaded byAlim Sheikh
- Simple FFT and Filtering Tutorial With Matlab - CodeProjectUploaded bycejavier
- 00 ELE 5 Orientation.pdfUploaded byJB Navarro
- Main PronectUploaded byanandbabugopathoti
- Filtro NotchUploaded byJose Leon
- Test 2 Sem 1 1415 SolutionUploaded bydaud
- active filter matlab final.docxUploaded byHakan Yamalı
- Digital Technologies for Controlling Room AcousticsUploaded byJosé Luis Duque Castro
- Discrete Time ConvolutionUploaded byjoydeep12
- 00083591Uploaded byquantumflight

- publist_sciex-1.pdfUploaded byNiranjanWangjam
- List of JournlsUploaded bytjdandin1
- CSL-TR-94-617Uploaded byمنال الكبيسي
- Fast FIR-Booth Algorithm.pdfUploaded byvvigya
- Sampling Rate ApplicationsUploaded byvvigya
- Samples MathematicsUploaded byvvigya
- w BksampTypesetUploaded byEverlin Martins
- 313042Uploaded byvvigya
- CIC2014 PaperUploaded byvvigya
- Effects of AliasingUploaded byvvigya
- Wp Digital if DplUploaded byvvigya
- CSVT93Uploaded byvvigya
- EJSR_49_4_15Uploaded byvvigya
- 10.1.1.131Uploaded byvvigya
- aravind_kean_reportUploaded byvvigya

- Tsunami Bulletin Number 018 Pacific Tsunami WarningUploaded byRodrigo Velasco
- Military Parade 5Uploaded byDragovic Slobodan
- Demantra Performance TuningUploaded bychandanjee
- DSEAR GuidanceUploaded byTitash Mukherjee Sen
- Operation manual Sauven 600Uploaded bypruebas
- 12.4.2012Uploaded byrskr_t
- Manual Nera Networks Evolution Xpand Revize CUploaded byariz0803
- KHUSHBOO AGARWAL.docxUploaded byLuckyGupta
- Gams BasicsUploaded byoidaadm
- Handbook of thermoUploaded byBogdan Tanasie
- MCM New ProductsUploaded bymerc2
- Hybrid ModulationUploaded byprabath jayasekara
- XXIV-Poster-14.pdfUploaded byArcangelo Di Tano
- Company Profile MHDLUploaded byYousuf Munni
- Irreversibility Analysis of Three Stage Vapour Compression Refrigeration Systems With Flash-Intercooler Using Eco-friendly New RefrigerantsUploaded byIjrei Journal
- cmiller_CSW_2010Uploaded bydty001
- Arms_2010-6Uploaded bymaruka33
- syllabusUploaded byapi-239204876
- SyllabusUploaded byrahul100100
- Mission Specification For A Jet Transport.docxUploaded byAdha Luri
- Eor Bulletin Sept 25 2013Uploaded byea
- MA4001_Sem1-15-16 - v1Uploaded byPearlyn Tiko Teo
- D-link Dsl g684tUploaded byIvan Sanader
- Blind People RoadcrossUploaded byKumarecit
- Cool Room TempUploaded byAditya Suardi
- MIMO Control DesignUploaded byKhoa Nguyen Dang
- Persistent Forecasting of Disruptive TechnologiesUploaded byJeff Begley
- b 954297Uploaded byhartmann100
- Transistor Tutorial About Bipolar and FET Transistors-nupUploaded bylosmilzo
- Mahindra Comviva - Digital Banking Caselet - 20Jul18 - RELEASE 1.1_cas_17Uploaded bysjiit007